Does the Effect of Pollution on Infant Mortality Differ Between Developing and Developed Countries? Evidence from Mexico City
Arceo-Gomez, Hanna and Oliva
APEC 8990
Paper Presentations
September 12, 2024
Motivation
- Air pollution is a significant issue in many parts of the world
- Studies have estimated the health effects of pollution in developed countries
- Reason to believe this evidence may not be externally valid for developing countries
- There may be a non-linear dose–response relationship between pollution and infant mortality
- The effect of pollution on health may be highly dependent on behavior
Research Question
What is the effect of pollution on infant mortality in the context of a developing country?
Overview
- Context: Mexico 1997 – 2006
- Rapidly industrializing during this time period
- Also implemented policies aimed at improving air quality
- Methods: fixed effects and IV using thermal inversions
- Findings:
- Evidence that pollution increases infant mortality
- Compared to the US, larger marginal effects for CO and similar marginal effects for PM10
- Concern: other factors may be correlated with health
- Weather, socio-economic status and changes in economic conditions
Empirical Strategy
Fixed Effects
Objective: estimate \beta_1 the relationship between pollution (P_{mw}) in a municipality (m) in a given week (w) and mortality per 100,000 live births (Y_{mw})
Y_{mw} = \beta_0 + \beta_1 P_{mw} + \alpha_m + \sigma_{mj} + \epsilon_{mw}
- \alpha_m is a set of municipality fixed effects that control for permanent differences across municipalities
- \sigma_{mj} is a set of bimonthly x municipality fixed effects, which control for common factors in a given two month block that could affect both pollution levels and infant mortality within a municipality
Empirical Strategy
Fixed Effects
Y_{mw} = \beta_0 + \beta_1 P_{mw} + \alpha_m + \sigma_{mj} + \epsilon_{mw}
Concerns:
- Unobserved, time-varying differences across municipalities will bias \beta_1
- Classical measurement error in the pollution variable will bias \beta_1 downwards
Empirical Strategy
Instrumental Variables
Objective: test whether inversions increase the concentrations of different types of pollutants \rightarrow use the number of thermal inversions in a given week (TI_w) to instrument for pollution
\begin{align*}
P_{mw} &= \pi_0 + \pi_1 TI_w + \sum \pi_{2m} w + h(W_{mw}) + \alpha_m + \sigma_{mj} + \mu_{mw} \\
Y_{mw} &= \beta_0 + \beta_1 P_{mw} + \sum \beta_{2m} w + h(W_{mw}) + \alpha_m + \sigma_{mj} + \epsilon_{mw}
\end{align*}
- TI_w varies at the week level so week by year fixed effects are not identified
- control for municipality-specific week by year trends (w)
- municipality fixed effects (\alpha_m) control for time-invariant characteristics across municipalities
- bimonthly x municipality fixed effects (\sigma_{mj}) control for seasonal effects within each municipality
- h(W_{mw}) is a set of controls for temperature and weather conditions
Empirical Strategy
Instrumental Variables
\begin{align*}
P_{mw} &= \pi_0 + \pi_1 TI_w + \sum \pi_{2m} w + h(W_{mw}) + \alpha_m + \sigma_{mj} + \mu_{mw} \\
Y_{mw} &= \beta_0 + \beta_1 P_{mw} + \sum \beta_{2m} w + h(W_{mw}) + \alpha_m + \sigma_{mj} + \epsilon_{mw}
\end{align*}
Concerns:
- Residual seasonal variation
- Short time frame (such as week) may overstate the effect
- Exclusion restriction may be violated
- Single instrument for all pollutants
Results
Instrument Relationship
![]()
Results
IV First Stage
Results
IV
Results
IV, Cause of Death
Results
IV, Multiple Pollutants
Results
Comparison with US
Summary
- Statistically significant effects of pollution on infant mortality
- Compared to US:
- Larger effects for CO
- Comparable estimates for PM10, despite much higher levels
- Policy decisions need to consider the different effects in developing countries
- May understate the benefits from environmental regulation